Rename ifm to ife for struct ifmedia_entry *. No functional change.

(msaitoh)

+ci_onproc

(skrll)

Fix macppc/dev/if_mc.c::mc_mediachange(). Not tested.

ifmedia_change() is used to change the device's media setting from
user-selected media. The user-selected media is not sc->sc_media.ifm_media but
sc->sc_media.ifm_cur->ifm_media. Note that mc_mediachange() is not called from
anywhere because it's disabled with #ifdef NOTYET.

(msaitoh)

Consistency.

(skrll)

Disable rngtest on output of cprng_strong.

We already do a self-test for correctenss of Hash_DRBG output;
applying rngtest to it does nothing but give everyone warning fatigue
about spurious rngtest failures.

(riastradh)

Fix admsw_mediachange(). Not tested.

ifmedia_change() is used to change the device's media setting from
user-selected media. The user-selected media is not
sc->sc_ifmedia[port]->ifm_media but sc->sc_ifmedia[port].ifm_cur->ifm_media.

(msaitoh)

PR/54730: Izumi Tsutsui: Use /var/db/obsolete/<set> to remove obsolete rc.d
files when not in $SOURCEMODE.

(christos)

Fix rump definition of cpu_number().

(riastradh)

Update rnd(9) man page to reflect reality since netbsd-7.

- Note rndsource_setcb, RND_FLAG_HASCB, and rnd_add_data_sync.
- Note user's obligation to serialize access to each rndsource.

(riastradh)

Fix macro error if QAT_DUMP is defined for debugging.

(hikaru)

Introspection. From "Surely You're Joking, Mr. Feynman!"

(sevan)

- Add some more failsafes to the CPU topology stuff, and build a 3rd
  circular list of peer CPUs in other packages, so we might scroll through
  them in the scheduler when looking to distribute or steal jobs.

- Fold the run queue data structure into spc_schedstate.  Makes kern_runq.c
  a far more pleasant place to work.

- Remove the code in sched_nextlwp() that tries to steal jobs from other
  CPUs.  It's not needed, because we do the very same thing in the idle LWP
  anyway.  Outside the VM system this was one of the the main causes of L3
  cache misses I saw during builds.  On my machine, this change yields a
  60%-70% drop in time on the "hackbench" benchmark (there's clearly a bit
  more going on here, but basically being less aggressive helps).

(ad)

build fix

(jmcneill)

dtrace: add support for aarch64

(jmcneill)

Define lwp_trapframe() macro

(jmcneill)

Add KDTRACE_HOOKS support.

(jmcneill)

Drop entries for "Scott Bartram and Frank van der Linden" and "the
University of Illinois at Urbana and their contributors".  They are
not referenced anywhere in the tree.

(uwe)

Add a comment to the entry for Computer Systems Laboratory at the
University of Utah.  It's missed by the extraction script b/c of the
non-standard formatting of the ad clause.

(uwe)

Move the entry for Caldera to its proper place in the sorting order.

(uwe)

Use troff \(xx names for the few accented characters we have here.

Extraction script outputs them literally, but some of them are in
utf-8 and some in latin1.  While here also use \(:o for Bodo Moeller
name as instructed.  Lule\(oa is now emited by the extraction script
directly.

This gives us the right characters in PostScript and HTML output.  In
text output they are transliterated with -mtty-char.

(uwe)

dm: Make global dm locks static

These are all per-file locks which protect dev/pdev/target structure.
taken-from: DragonFlyBSD

(tkusumi)

Translate Lule} to troff's Lule\(oa.

(uwe)

dm: Remove unused macro MAX_TARGET_STRING_LEN

that was added but never used in both NetBSD and DragonFlyBSD.
taken-from: DragonFlyBSD

(tkusumi)

dm: Include <sys/vnode.h> in dm.h

dm.h already depends on vnode, so have dm.h include <sys/vnode.h>
instead of other .c files.

taken-from: DragonFlyBSD

(tkusumi)

dm: Doesn't need to expose struct cmd_function in a header

(tkusumi)

Use __insn_barrier to enforce ordering in l_ncsw loops.

(Only need ordering observable by interruption, not by other CPUs.)

(riastradh)

dm: Remove unused macro DM_CMD_LEN

that was added but never used in both NetBSD and DragonFlyBSD.
taken-from: DragonFlyBSD

(tkusumi)

Use groff -Tencoding argument to emit <meta charset="..."/>

(uwe)

Calulate the right lenght for isoc transfer and the usb_syncmem

PR/54696: Kernel panic in bus_dma.c on Raspberry Pi 3B/3B+

(skrll)

Use size_t for loop count variable

(skrll)

Adapt to use the new atomic_* operations.  As discussed with
riastradh@ in private Email.

(pgoyette)

Add xsrc to the invocation example.

(uwe)

Drop the final quote if followed by dot.

(uwe)

Need <sys/sysctl.h> for previous

(skrll)

Stopgap hack to unbreak the build: #ifdef __HAVE_ATOMIC64_LOADSTORE
the event counter update. From rmind@

(martin)

One more tweak.

(ad)

tlb_exception: in slow path, check for usermode before calling userret().

(ad)

Make sure the assignment to "idepth" is done inside the loop to prevent
preemption between loop end and dereference of "l_cpu->ci_depth".

(hannken)

Make cpu_intr_p() work with "curlwp->l_cpu == NULL" and
assert "curlwp == &lwp0" in this case.

Prevents crash during early boot with "options LOCKDEBUG".

(hannken)

streamline receive filter logic, work-in-progress.

(nisimura)

Make dwc2debug sysctl'able

(skrll)

Apply a fix from upstream:
  https://github.com/NLnetLabs/unbound/pull/122
This should enable proper functioning of tcp-idle-timeout.

(he)

KNF

(skrll)

Remove accidentally committed output from extract-contrib-string.pl
I will have a stern talk with my emacs about this incident.

(uwe)

Fix the enry for Computer Systems Laboratory at the University of Utah
and move it to its proper place in the sorting order.

(uwe)

Move the entry for CMU Computing Services to its proper place in the
sorting order.  Add more ad clauses from CMU ppp code.

(uwe)

Move the entry for CMU Computing Services to its proper place in the
sorting order.

(uwe)

Remove FDT_INTR_MPSAFE flag.

(hkenken)

Recognize ad clause in CMU PPP and some OpenSSL code.

Use "n" (grouping doesn't capture) when matching $ack* regexps to
alleviate performance hit a bit.

(uwe)

Rip out pserialize(9) logic now that the RCU patent has expired.

pserialize_perform() is now basically just xc_barrier(XC_HIGHPRI).
No more tentacles throughout the scheduler.  Simplify the psz read
count for diagnostic assertions by putting it unconditionally into
cpu_info.

From rmind@, tidied up by me.

(riastradh)

Fix inequality for refcnt drain: -1 here means all refs gone.

(riastradh)

Fix order of nulling un->un_pri->unp_ec.ec_mii.

Can't null it until after if_detach prevents further use.

While here, fix conditionals in usbnet_tick_task to use the unp_dying
flag, not the nullness of mii (or of ifp, which never null because
it's an embedded member).

(riastradh)

Omit obsolete comment and needless __diagused marker.

(riastradh)

Fix brain fart in alignment criterion.

(riastradh)

Skip fences in bus_space_barrier on I/O space.

I/O operations are issued in program order.  Not that I/O operations
are usually a performance bottleneck anyway, but maybe it is slightly
cheaper to avoid stalling on store buffers or pending loads, and
there's very little cost to the skipping criterion here.

(riastradh)

Add HMAC-SHA-{256,384,512} test cases.

(hikaru)

s/upate/update/ in comment.

(msaitoh)

Special-case badly worded ad clause in src/sys/lib/libkern/rngtest.c

(uwe)

This product includes software developed by Ravikanth.

(uwe)

Fix typo in regexp to detect m4 comments.

(uwe)

Fix typo in ad clause.  Ok perry@.

(uwe)

Swap Per Fogelstrom entries.  Missed in previous.

(uwe)

Detect more ad clauses, pick more duplicates.

If there are notices that differ in the final dot, prefer the version
with the dot.  Sort "by the" enries after all other "by Name" entries.
With the corresponding re-ordering of the legal.common the output of
this script is getting useful again.

(uwe)

Sort entries "by Foo." before "by Foo and" etc.  Sort the section with
"for the NetBSD" notices that had a lot of entries out of order.  Drop
final dots that are not in any of the required notices to reduce diff
with the script output.

(uwe)

Move entries with short first line that ends with "by" to their proper
places in the sorting order.

(uwe)

Restore entries for "K. Kobayashi" and "K. Kobayashi and H. Shimokawa".
The script missed them as it was confused by the slighly non-standard
license wording.  My apologies.

(uwe)

Restore the entry for Yasushi Yamasaki - it comes from xsrc,
my apologies.

(uwe)

Take the basic CPU topology information we already collect, and use it
to make circular lists of CPU siblings in the same core, and in the
same package.  Nothing fancy, just enough to have a bit of fun in the
scheduler trying out different tactics.

(ad)

Drop the entry for "University of California, Berkeley and its
contributors, as well as the Trustees of Columbia University" that was
required by the ad clause in external/bsd/am-utils/dist/COPYING.
The ad clause was dropped in the versions we have since netbsd-8.

(uwe)

Don't use an undefined variable.

(uwe)

Fix s/product/model/ search-and-destroy accident that affected license.

Back in 1998 thorpej@ adapted this from one of the several versions we
have in the tree but needed s/product/model/ in the text of the
script.  Unfortunately that also affected the license text: "This
model includes software..." and "...or promote models".

(uwe)

Fix tyo in comment

(pgoyette)

Drop the entry for Yasushi Yamasaki.  Nothing in the tree requires it.

(uwe)

Move the entry for Kyma Systems to its proper place in the sorting order.

(uwe)

Drop the entry for David Muir Sharnoff.  It was for expn(1) script in
am-utils that was never installed and is no longer there since netbsd-8.

(uwe)

Drop the entry for Thomas Gerner.

thomas@ dropped clauses 3 and 4 from his license back in 2009 and the
entry was dropped from here, but a copy was forgotten in the atari
port-specific legal file and merged back here in 2015.

(uwe)

fix the build for when UVMHIST is enabled.

(chs)

Fix a hard hang with Xen MP.

(ad)

Drop the entry for Texas A&M University.
Nothing in the tree has its ad clause.

(uwe)

Drop advertisement clause from my licenses

(bouyer)

Move the entry for Matthew Fredette to its proper place in the sorting order.

(uwe)

drop my name from the advertising clause.

(christos)

drop my name from the advertising clause.

(christos)

Define PT_GETXMMREGS and PT_SETXMMREGS in PT_MACHDEP_STRINGS/amd64

(kamil)

+ ci_onproc

(ad)

Drop entries for "K. Kobayashi" and "K. Kobayashi and H. Shimokawa"
that are no longer referenced in tree.

(uwe)

Add entry for Cisco for the SCTP code.

(uwe)

Add entry for Ben Gray.

(uwe)

Leave it entirely up to the driver whether to map prefetchable.

We have only a few drivers that pass BUS_SPACE_MAP_PREFETCHABLE to
pci_mapreg_map -- dev/pci/if_hme_pci.c, dev/pci/igma.c,
dev/pci/radeonfb.c, dev/pci/wcfb.c -- and they all do it for ROM or
framebuffers.

Based on a subthread from macallan@ in the earlier discussion:

https://mail-index.NetBSD.org/tech-kern/2017/03/23/msg021685.html

The thrust is that:
- a driver not asking for prefetchable mappings shouldn't get it
- some devices are correctly used with prefetchable mappings, but for
  some reason fail to set the prefetchable bit in the BAR
- nobody could identify any classes of device for which
  (a) the driver asks for prefetchable mappings, but
  (b) certain matching devices can't actually be used with
      prefetchable mappings

This brings pci_mapreg_map in line with what the documentation says;
the documentation never advertised that the prefetchable bit in the
BAR could cause the bus_space mapping to be prefetchable.

(riastradh)

Fix ISO-646-SE spelling of Lule\[oa].

(uwe)

Collapse multiple spaces in an entry (doesn't affect sorting).

(uwe)

Move the entry for Mika Kortelainen to its proper place in the sorting
order after fixing multiple spaces.

(uwe)

Collapse multiple spaces between words.
There are a few entries with "by__Name" that affects sorting.

(uwe)

Move Intel's entry to its proper place in the sorting order.

(uwe)

dm: Remove misleading comment on linear target arg

The offset arg is mandatory.
Remove code and comment that makes it sounds like it's optional.

taken-from: DragonFlyBSD

(tkusumi)

dm: Add a comment on race window on unload

There is a minor race window on unload vs device creation
that can cause panic.
https://github.com/DragonFlyBSD/DragonFlyBSD/commit/53a07f3ae7313aa58948a60f46428bfc2254dc3c

taken-from: DragonFlyBSD

(tkusumi)

remove duplicate word

(martin)

Ammend dedication

(martin)

Add dedication to Eric Schnoebelen.

(martin)

postfix-3.4.8 out.

(wiz)

Use LFENCE/SFENCE/MFENCE in x86 bus_space_barrier.

These are needed for BUS_SPACE_MAP_PREFETCHABLE mappings.  On x86,
these are WC-type memory regions, which means -- unlike normal
WB-type memory regions -- loads can be reordered with loads,
requiring LFENCE, and stores can be reordered with stores, requiring
SFENCE.

Reference: AMD64 Architecture Programmer's Manual, Volume 2: System
Programming, Sec. 7.4.1 `Memory Barrier Interaction with Memory
Types', Table 7-3 `Memory Access Ordering Rules'.

(riastradh)

Use BUS_SPACE_MAP_PREFETCHABLE only if BAR and driver agree on it.

- A driver that expects prefetchable memory and knows to issue the
  needed bus_space_barrier calls can pass BUS_SPACE_MAP_PREFETCHABLE
  to indicate a desire to map the memory prefetchable if the BAR
  allows it.

  (A driver that _really wants_ BUS_SPACE_MAP_PREFETCHABLE even if
  the BAR claims _not_ to be prefetchable can use pci_mapreg_info and
  bus_space_map explicitly -- this is not different from what we have
  today.)

- For a driver that _does not_ expect prefetchable memory, the
  appearance of the prefetchable bit in the BAR shouldn't cause it to
  use BUS_SPACE_MAP_PREFETCHABLE, because the driver will not issue
  the needed bus_space_barrier calls to get sensible results.

Note: `Prefetchable' here, sometimes called `write-combining', means
reads have no side effects, and writes are idempotent, so it is safe
to issue reads out of order and safe to combine writes.

Mappings with BUS_SPACE_MAP_PREFETCHABLE are often more weakly
ordered than normal memory -- e.g., on x86, in WC-type memory
regions, loads can be reordered with loads, stores can be reordered
with stores, which is not possible with any other type of memory
regions.

Discussed on tech-kern a while ago:

https://mail-index.NetBSD.org/tech-kern/2017/03/22/msg021678.html

This is option A, which received the most support.  This should help
unconfuse drivers that do not expect prefetchable mappings, like
Yamaguchi-san tripped over recently:

https://mail-index.NetBSD.org/tech-kern/2019/12/02/msg025785.html

(riastradh)

Use PCI_MSIX_"TBL"BIR_MASK instead of PCI_MSIX_"PBA"BIR_MASK for MSI-X table.
This is not a real bug because both macros have the same value.

(msaitoh)

Add cfi annotations so that gdb can unwind the stack through signal handlers.

(christos)

Add more tests for variable modifiers in make.

(rillig)

Fix out-of-bounds read in Str_Match.

(rillig)

Add missing #include <sys/atomic.h>

(uwe)

SCSI OSD

(sevan)

Don't deregister twice with pmf.

(mlelstv)

Reset MCU ready status before resetting the MCU.
Fixes PR kern/54728

(mlelstv)

Attempt to load the zfs module even if /etc/zfs/zpool.cache is absent. The
module needs to be loaded to create a pool in the first place, and
autoloading won't work after the fact won't work at securelevel=1.

(jmcneill)

Restore xcall(9) fast path using atomic_load/store_*.

While here, fix a bug that was formerly in xcall(9): a missing
acquire operation in the xc_wait fast path so that all memory
operations in the xcall on remote CPUs will happen before any memory
operations on the issuing CPU after xc_wait returns.

All stores of xc->xc_donep are done with atomic_store_release so that
we can safely use atomic_load_acquire to read it outside the lock.
However, this fast path only works on platforms with cheap 64-bit
atomic load/store, so conditionalize it on __HAVE_ATOMIC64_LOADSTORE.
(Under the lock, no need for atomic loads since nobody else will be
issuing stores.)

For review, here's the relevant diff from the old version of the fast
path, from before it was removed and some other things changed in the
file:

diff --git a/sys/kern/subr_xcall.c b/sys/kern/subr_xcall.c
index 45a877aa90e0..b6bfb6455291 100644
--- a/sys/kern/subr_xcall.c
+++ b/sys/kern/subr_xcall.c
@@ -84,6 +84,7 @@ __KERNEL_RCSID(0, "$NetBSD: subr_xcall.c,v 1.27 2019/10/06 15:11:17 uwe Exp $");
#include <sys/evcnt.h>
#include <sys/kthread.h>
#include <sys/cpu.h>
+#include <sys/atomic.h>

#ifdef _RUMPKERNEL
#include "rump_private.h"
@@ -334,10 +353,12 @@ xc_wait(uint64_t where)
xc = &xc_low_pri;
}

+#ifdef __HAVE_ATOMIC64_LOADSTORE
/* Fast path, if already done. */
- if (xc->xc_donep >= where) {
+ if (atomic_load_acquire(&xc->xc_donep) >= where) {
return;
}
+#endif

/* Slow path: block until awoken. */
mutex_enter(&xc->xc_lock);
@@ -422,7 +443,11 @@ xc_thread(void *cookie)
(*func)(arg1, arg2);

mutex_enter(&xc->xc_lock);
+#ifdef __HAVE_ATOMIC64_LOADSTORE
+ atomic_store_release(&xc->xc_donep, xc->xc_donep + 1);
+#else
xc->xc_donep++;
+#endif
}
/* NOTREACHED */
}
@@ -462,7 +487,6 @@ xc__highpri_intr(void *dummy)
* Lock-less fetch of function and its arguments.
* Safe since it cannot change at this point.
*/
- KASSERT(xc->xc_donep < xc->xc_headp);
func = xc->xc_func;
arg1 = xc->xc_arg1;
arg2 = xc->xc_arg2;
@@ -475,7 +499,13 @@ xc__highpri_intr(void *dummy)
* cross-call has been processed - notify waiters, if any.
*/
mutex_enter(&xc->xc_lock);
- if (++xc->xc_donep == xc->xc_headp) {
+ KASSERT(xc->xc_donep < xc->xc_headp);
+#ifdef __HAVE_ATOMIC64_LOADSTORE
+ atomic_store_release(&xc->xc_donep, xc->xc_donep + 1);
+#else
+ xc->xc_donep++;
+#endif
+ if (xc->xc_donep == xc->xc_headp) {
cv_broadcast(&xc->xc_busy);
}
mutex_exit(&xc->xc_lock);

(riastradh)

Avoid calling pmap_page_protect() while under uvm_pageqlock.

(ad)

Enable ZFS support on aarch64

(jmcneill)

Flush insn / data caches after loading modules

(jmcneill)

Need sys/atomic.h on NetBSD

(jmcneill)

Provide a default ptob() implementation

(jmcneill)

Initialize b_dev before passing buf to d_minphys (ldminphys needs this)

(jmcneill)

Build aarch64 modules without fp or simd instructions.

(jmcneill)

From 02c3dad0f3b4d26e0faa5cc51d06bc50d693dcdc Mon Sep 17 00:00:00 2001
From: Brian Paul <brianp@vmware.com>
Date: Wed, 9 Oct 2019 12:05:16 -0600
Subject: [PATCH] Call shmget() with permission 0600 instead of 0777

A security advisory (TALOS-2019-0857/CVE-2019-5068) found that
creating shared memory regions with permission mode 0777 could allow
any user to access that memory.  Several Mesa drivers use shared-
memory XImages to implement back buffers for improved performance.

This path changes the shmget() calls to use 0600 (user r/w).

Tested with legacy Xlib driver and llvmpipe.

Cc: mesa-stable@lists.freedesktop.org
Reviewed-by: Kristian H. Kristensen <hoegsberg@google.com>

(maya)

Another instance of cpu_onproc to replace.

(ad)

Regen.

(ad)

Back out previous temporarily - seeing unusual lookup failures.  Will
come back to it.

(ad)

Add ci_onproc.

(ad)

cpu_onproc -> ci_onproc

(ad)

Switch in_interrupt() in KCOV to cpu_intr_p()

This makes KCOV more MI friendly and removes x86-specific in_interrupt()
implementation.

(kamil)

Disable KCOV instrumentation in x86_machdep.c

This allows to use cpu_intr_p() directly inside KCOV.

(kamil)

Init kern_runq and kern_synch before booting secondary CPUs.

(ad)

Back out the fastpath change in xc_wait().  It's going to be done differently.

(ad)

Free pages in batch instead of taking uvm_pageqlock for each one.

(ad)

__cacheline_aligned on a lock.

(ad)

NetBSD 9.99.19 - many kernel data structure changes

(ad)

dm: Fix race on pdev create

List lookup and insert need to be atomic.
https://github.com/DragonFlyBSD/DragonFlyBSD/commit/ac816675c8321309b49d6a335f95c5388036803f

take-from: DragonFlyBSD

(tkusumi)

Make the fast path in xc_wait() depend on _LP64 for now.  Needs 64-bit
load/store.  To be revisited.

(ad)

Mark unreachable branch with __unreachable() to fix i386/ALL build.

(riastradh)

Fix false sharing problems with cpu_info.  Identified with tprof(8).
This was a very nice win in my tests on a 48 CPU box.

- Reorganise cpu_data slightly according to usage.
- Put cpu_onproc into struct cpu_info alongside ci_curlwp (now is ci_onproc).
- On x86, put some items in their own cache lines according to usage, like
  the IPI bitmask and ci_want_resched.

(ad)

Adapt <sys/pslist.h> to use atomic_load/store_*.

Changes:

- membar_producer();
  *p = v;

    =>

  atomic_store_release(p, v);

  (Effectively like using membar_exit instead of membar_producer,
  which is what we should have been doing all along so that stores by
  the `reader' can't affect earlier loads by the writer, such as
  KASSERT(p->refcnt == 0) in the writer and atomic_inc(&p->refcnt) in
  the reader.)

- p = *pp;
  if (p != NULL) membar_datadep_consumer();

    =>

  p = atomic_load_consume(pp);

  (Only makes a difference on DEC Alpha.  As long as lists generally
  have at least one element, this is not likely to make a big
  difference, and keeps the code simpler and clearer.)

No other functional change intended.  While here, annotate each
synchronizing load and store with its counterpart in a comment.

(riastradh)

Rework modified atomic_load/store_* to work on const pointers.

(riastradh)

Fix a longstanding problem with LWP limits.  When changing the user's
LWP count, we must use the process credentials because that's what
the accounting entity is tied to.

Reported-by: syzbot+d193266676f635661c62@syzkaller.appspotmail.com

(ad)

Remove the pretty much useless 128MB swap partition from the arm images.

(jmcneill)

Make cpu_intr_p() safe to use anywhere, i.e. outside assertions:

Don't call kpreempt_disable() / kpreempt_enable() to make sure we're not
preempted while using the value of curcpu().  Instead, observe the value of
l_ncsw before and after the check to see if we have been preempted.  If
we have been preempted, then we need to retry the read.

(ad)

Minor correction to previous.

(ad)

- Adjust uvmexp.swpgonly with atomics, and make uvm_swap_data_lock static.
- A bit more __cacheline_aligned on mutexes.

(ad)

Deactivate pages in batch instead of acquiring uvm_pageqlock repeatedly.

(ad)

Give each of the page queue locks their own cache line.

(ad)

Activate pages in batch instead of acquring uvm_pageqlock a zillion times.

(ad)

If the system is not up and running yet, just run the function locally.

(ad)

Map the video RAM cacheable/prefetchable, it's very slow and this helps a bit.

(ad)

Update to match change in layout of vnode LRU lists.

(ad)

Regen for VOP_LOCK & LK_UPGRADE/LK_DOWNGRADE.

(ad)

Minor vnode locking changes:

- Stop using atomics to maniupulate v_usecount.  It was a mistake to begin
  with.  It doesn't work as intended unless the XLOCK bit is incorporated in
  v_usecount and we don't have that any more.  When I introduced this 10+
  years ago it was to reduce pressure on v_interlock but it doesn't do that,
  it just makes stuff disappear from lockstat output and introduces problems
  elsewhere.  We could do atomic usecounts on vnodes but there has to be a
  well thought out scheme.

- Resurrect LK_UPGRADE/LK_DOWNGRADE which will be needed to work effectively
  when there is increased use of shared locks on vnodes.

- Allocate the vnode lock using rw_obj_alloc() to reduce false sharing of
  struct vnode.

- Put all of the LRU lists into a single cache line, and do not requeue a
  vnode if it's already on the correct list and was requeued recently (less
  than a second ago).

Kernel build before and after:

119.63s real  1453.16s user  2742.57s system
115.29s real  1401.52s user  2690.94s system

(ad)

Regen.

(ad)

Make nc_hittime volatile to defeat compiler cleverness.

(ad)

namecache changes:

- Delete the per-entry lock, and borrow the associated vnode's v_interlock
  instead.  We need to acquire it during lookup anyway.  We can revisit this
  in the future but for now it's a stepping stone, and works within the
  quite limited context of what we have (BSD namecache/lookup design).

- Implement an idea that Mateusz Guzik (mjg@FreeBSD.org) gave me.  In
  cache_reclaim(), we don't need to lock out all of the CPUs to garbage
  collect entries.  All we need to do is observe their locks unheld at least
  once: then we know they are not in the critical section, and no longer
  have visibility of the entries about to be garbage collected.

- The above makes it safe for sysctl to take only namecache_lock to get stats,
  and we can remove all the crap dealing with per-CPU locks.

- For lockstat, make namecache_lock a static now we have __cacheline_aligned.

- Avoid false sharing - don't write back to nc_hittime unless it has changed.
  Put a a comment in place explaining this.  Pretty sure this was there in
  2008/2009 but someone removed it (understandably, the code looks weird).

- Use a mutex to protect the garbage collection queue instead of atomics, and
  adjust the low water mark up so that cache_reclaim() isn't doing so much
  work at once.

(ad)

PR port-sparc/54718 (sparc install hangs since recent scheduler changes)

- sched_tick: cpu_need_resched is no longer the correct thing to do here.
  All we need to do is OR the request into the local ci_want_resched.

- sched_resched_cpu: we need to set RESCHED_UPREEMPT even on softint LWPs,
  especially in the !__HAVE_FAST_SOFTINTS case, because the LWP with the
  LP_INTR flag could be running via softint_overlay() - i.e. it has been
  temporarily borrowed from a user process, and it needs to notice the
  resched after it has stopped running softints.

(ad)

minor adjustments, to avoid warnings on debug builds

(maxv)

sh3: make ASTs work as expected, and fix a few things in the TLB refill path.
With help from uwe@ and martin@.

(ad)

Note added dedication

(martin)

Add dedication for Matthias Drochner.

(martin)

Add missing <sys/atomic.h> include

(martin)

localify

(maxv)

localify

(maxv)

Use atomic_{load,store}_relaxed() on global counters.

(maxv)

Use unsigned to avoid undefined behavoir. Found by kUBSan.

(msaitoh)

Add KCSAN instrumentation for atomic_{load,store}_*.

(maxv)

dm: Remove unused dm_dev::dev_type

Given OOP-like architecture of dm target device structure,
dm_dev doesn't need to have self contained target type field,
and in fact this is unused.

(tkusumi)

Make owtemp reliable for me:

- Don't do the calculation if there is a CRC error.
- If we get any kind of error during a refresh, retry up to three times.
- Add event counters to report what's going on.

(ad)